Sendbird · Rate Limits

Sendbird Rate Limits

Sendbird Platform API enforces tier-based rate limits keyed to the application's MAU plan, with separate ceilings for GET vs POST/PUT/DELETE. Specific endpoint families (message migration, user updates, channel creation, invitations, message sends, mark-as-read) are additionally capped at 5 requests/sec regardless of tier. Limits are signaled via X-RateLimit-* headers; throttled requests return HTTP 429 with code 500910.

Sendbird Rate Limits is the machine-readable rate-limit profile for Sendbird on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 13 rate-limit definitions, measuring requests_per_second.

The profile also includes 4 backoff/retry policies defined and response codes documented for throttled.

Tagged areas include Chat, Messaging, and Rate Limiting.

13 Limits Throttle: 429
ChatMessagingRate Limiting

Limits

Free trial — GET application
requests_per_second · second
10
Free trial — POST/PUT/DELETE application
requests_per_second · second
5
5K MAU — GET application
requests_per_second · second
60
5K MAU — POST/PUT/DELETE application
requests_per_second · second
10
10K MAU — GET application
requests_per_second · second
100
10K MAU — POST/PUT/DELETE application
requests_per_second · second
20
25K MAU — GET application
requests_per_second · second
200
25K MAU — POST/PUT/DELETE application
requests_per_second · second
40
50K MAU — GET application
requests_per_second · second
400
50K MAU — POST/PUT/DELETE application
requests_per_second · second
100
100K+ MAU — GET application
requests_per_second · second
600
100K+ MAU — POST/PUT/DELETE application
requests_per_second · second
200
Stricter endpoints (all tiers) application/endpoint
requests_per_second · second
5
Applies to message migration, user updates (PUT /users/{user_id}), group channel creation, invitations, message sends, and mark-as-read endpoints regardless of plan tier.

Policies

HTTP method tiering
GET endpoints get 2–6× the throughput of mutating methods at the same MAU tier; design reads and writes accordingly.
Endpoint-specific stricter cap
A short list of write-heavy endpoints (channel creation, invitations, message sends, mark-as-read, user updates, migration) are pinned at 5/sec irrespective of plan; bulk operations should be queued.
Backoff
On 429 responses, honor X-RateLimit-RetryAfter (seconds) and apply exponential backoff with jitter.
Window signaling
X-RateLimit-Limit is published in the form "N;w=W" (e.g., "100;w=60"); X-RateLimit-Remaining and X-RateLimit-Reset reflect the current window.

Sources